Machine Learning in 2D & 3D Data Clustering

In this course, I tackled image segmentation and 3D point cloud data clustering and visualization, exploring machine learning application in data analysis. 

The diagram below shows K means applied on 2D data, with iterative step 1 - 6. Reached ~95% accuracy in 6 steps. 

Applying K means to image data, the image can be segmented into various color clusters. 

Then Gaussian Mixture Models (GMMs) is built to handle more complicated datasets. Expectation-Maximization (EM) algorithm is used to iteratively refine the models. Applied GMMs to segment 3D point cloud data, the model is able to separate objects, backgrounds, and other structures effectively.